When one hears the word “robot,” the reaction frequently is one of either concern or amusement – concern because of movies like “The Terminator” and “I, Robot,” where machines decide it’s time for their makers to go, and amusement because of movies like “Star Wars” and “Wall-E” and their endearing, Chaplinesque metal characters. But this vision of robots is skewed, because movies are focused on entertainment rather than information. The real story of robots is much more complicated and remarkable.
Elliott Masie, president and CEO of the MASIE Center, recently said, “There is a growing belief that we are beginning to enter an era where the technology base [will] increasingly be able to understand how we learn.” LearnBots will be at the forefront of this new era.
What are LearnBots?
LearnBots are routines programmed to automate learning. For example, a LearnBot could track your learning history, map it against your current learning needs, provide you with the learning interventions you need, and even suggest options to increase your abilities in related knowledge or skill areas. Fairly sophisticated support of this type is currently available in the marketplace (think Amazon), and the learning industry is slowly adopting some of these benefits. Over time, however, LearnBots will be able to do much, much more.
Where did they come from?
LearnBots have strong ties to Electronic Performance Support Systems (EPSSs), which emerged in the late 1980s/early 1990s. EPSSs are systems that strive to minimize if not eliminate the distinction between work and learning. Initially, EPSSs focused on human programmed context-sensitive support. Now, more of this functionality is automated; Bob Mosher, chief learning evangelist at APPLY Synergies, observes that “these technologies will only get smarter and more adaptive as time goes on.”
There are also important ties to Intelligent Tutoring. Clark Quinn, executive director of Quinnovation, notes, “Intelligent Tutoring Systems have been around for decades, adapting on the basis of deep models of expertise in a domain. More recently, adaptation[s] based upon characteristics of the learner have emerged, and now some commercial ones have hit the market that combine some of both elements.”
But these adaptations have been slow to catch on. Clive Shepherd, founding director of The More Than Blended Learning Company, observes, “For some reason, we have restricted our use of computer[s] to mere multimedia distribution devices and completely ignored their potential for providing an adaptive learning experience,” even though “real progress was being made with Intelligent Tutoring” in the 1980s. Fortunately, he says, “…if you wait long enough, it all comes [a]round again, and this time, hopefully it will stick.”
LearnBots have benefited from and been influenced by developments in many fields, including:
- Personalization/Adaptive Learning
- Increases in Computing Power, which, in turn, lead to increased capability
- Sensors and the Internet of Everything (IOE): information gathered from everywhere and updated continuously
- Big Data and Predictive Analytics: the massive collection of information gathered from sensors and the IOE and the programs that use this information to predict outcomes
- Synthetic Biology (including emotional interfaces): reality mimicked in an electronic/synthetic form, which also serves as the interface to reality
Given their complexity, these and other developments in business, science, academia and the world in general will shape the various generations of LearnBots in ways that may be surprising.
Why LearnBots?
LearnBots can:
- Automate basic, repetitive and/or error-prone learning development tasks, freeing us to do other (presumably higher-order) activities
- Gather and interpret data to better manage learning administration
- Fill learning gaps by providing high-quality learning to more people who need it
- Provide efficiencies that will reduce the complexity, time and/or number of steps needed to reach learning goals
- Improve the quality of learning by providing appropriate consistency while delivering personalization
- Provide learning functionality and support at the speed of computer processing power
- Deliver adaptive learning using predictive analytics, semantic search engine optimization and other advanced techniques
- Be available continuously
In the future, LearnBots may also:
- Teach themselves, so they can continually improve
- Provide realistic, highly engaging and ever-changing simulations and other learning events designed to lead participants to new insights, opportunities and solutions
- Expand the definition of learning
- Expand the impact of learning
- Help us think differently about the world, which may help us make the world better
Will they change over time?
Personalization, adaptive learning, increases in computing power, sensors and other phenomena have shaped First Generation LearnBots – the ones available today. These LearnBots can make adjustments to an EPSS or help a Learning Management System (LMS) perform better. This generation will create new standards of learning efficiency, but they are just the beginning.
Building on what we learned from the First Generation and applying advances in mathematics, data science, psychology and other fields, Second Generation LearnBots will be able to understand, if not master, subtleties of human personality and, in particular, natural language. They will use this knowledge to drive and continually enhance deep and engaging simulations that will test human potential.
But there will be limits. The Second Generation LearnBot will still operate behind the scenes, the product of the intelligent machine language developed by scientists but continually enhanced by the machines themselves. They will raise the bar for effectiveness, but there will still be more to come.
Third Generation LearnBots will assume physical form as synthetic teachers. They will have the remarkable ability to understand and emphasize with their students, adapting their actions to provide unparalleled learning experiences. These LearnBots will employ an advanced form of heuristics, which will produce profound and novel insights that, in turn, will be transformed into unprecedented learning experiences. These experiences could alter the way people see the world and how they live in it. If LearnBots reach their potential, the impact will be transformative.
Why aren’t they prevalent now?
Julie Dirksen, learning strategy consultant at Usable Learning, says, “I’ve been puzzled by how little use the learning field has made of recommender engines. They are being used more and more often in retail, social media, and online content. It seems like a natural fit for learning environments, but I’ve seen very little adoption thus far. It probably has something to do with how locked down learning content tends to be, and our ways of managing learning content will need to evolve.”
Besides the challenges involved in handling learning content, there are at least two other reasons why LearnBots aren’t more prevalent.
First, successful LearnBot development requires the active collaboration of two separate and often very different disciplines: learning and data science. There is a precedent for the type of interdisciplinary cooperation needed (i.e., the work done in learning technologies over several decades). However, to date, work on LearnBots has been isolated and limited. The two groups attend different schools, work in different industries and have their own ways of speaking. We need a group of LearnBot scientists with the right academic and professional opportunities to develop this interdisciplinary role, and given the LearnBot’s potential, this profession is likely to develop.
Second, creating LearnBots is difficult. They support a key function of the human brain and, as physicist Dr. Michio Kaku points out, the human brain “is the most complex object in the known universe.” The mathematics needed to enable effective learning through robotics is a work in progress and may stay that way for some time. In addition, many of the disciplines that need to converge to produce Third or even Second Generation LearnBots are just emerging. Therefore, it may be many years before LearnBots think, look and act like us.
What is the future for LearnBots?
Craig Weiss, e-learning CEO, blogger, speaker and analyst, is cautious: “I think it will depend on whether or not training and the L&D departments see value in what [LearnBots] can do and, more importantly, the level [to] which…the administrator will be able to tweak the weighting or percentiles of the algorithm. If the LearnBot does not allow for any modifications on behalf of the organization, it will be a problem.”
Some of Weiss’ colleagues are more optimistic. Clark Quinn says, “The growing potential for semantic mining of content offers up the promise of systems that can meet contextualized learning needs without human intervention, though a hybrid solution is likely to be better. Still, we are already seeing such emergence, and the notion of a ubiquitous always-on learning mentor could reach fruition within [the next five to ten years].” Ryan Tracey, e-learning manager at AMP, agrees that LearnBots will likely become more prevalent but cautions that “if you need a Ph.D. in electronic engineering…the technology will remain within the realm of the few who can afford to pay someone else big money to do it for them.”
Because of the potential of LearnBots to save time, money and effort while continually improving their in-demand services, business will drive science and the learning profession to find a way to make them work. Will it happen in the next three years? Five years? Ten years? Stay tuned!